Date:   Fri, 4 Feb 2022 00:13:55 +0100
From:   Marijn Suijten <marijn.suijten@...ainline.org>
To:     Bjorn Andersson <bjorn.andersson@...aro.org>
Cc:     Pavel Machek <pavel@....cz>, Rob Herring <robh+dt@...nel.org>,
        Andy Gross <agross@...nel.org>,
        Thierry Reding <thierry.reding@...il.com>,
        Uwe Kleine-K?nig <u.kleine-koenig@...gutronix.de>,
        Lee Jones <lee.jones@...aro.org>, linux-leds@...r.kernel.org,
        devicetree@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-arm-msm@...r.kernel.org, linux-pwm@...r.kernel.org,
        Yassine Oudjana <y.oudjana@...tonmail.com>,
        Luca Weiss <luca@...tu.xyz>,
        Subbaraman Narayanamurthy <subbaram@...eaurora.org>
Subject: Re: [PATCH v10 2/2] leds: Add driver for Qualcomm LPG

On 2022-02-02 13:56:10, Bjorn Andersson wrote:
> On Wed 02 Feb 03:03 PST 2022, Marijn Suijten wrote:
> 
> > On 2022-01-28 18:50:42, Bjorn Andersson wrote:
> > > On Wed 27 Oct 16:19 CDT 2021, Marijn Suijten wrote:
> > > 
> > > > Hi Bjorn,
> > > > 
> > > > On 2021-10-22 10:25:35, Bjorn Andersson wrote:
> > > > > On Sat 09 Oct 21:39 PDT 2021, Bjorn Andersson wrote:
> > > > > 
> > > > > > The Light Pulse Generator (LPG) is a PWM-block found in a wide range of
> > > > > > PMICs from Qualcomm. These PMICs typically comes with 1-8 LPG instances,
> > > > > > with their output being routed to various other components, such as
> > > > > > current sinks or GPIOs.
> > > > > > 
> > > > > > Each LPG instance can operate on fixed parameters or based on a shared
> > > > > > lookup-table, altering the duty cycle over time. This provides the means
> > > > > > for hardware assisted transitions of LED brightness.
> > > > > > 
> > > > > > A typical use case for the fixed parameter mode is to drive a PWM
> > > > > > backlight control signal, the driver therefor allows each LPG instance
> > > > > > to be exposed to the kernel either through the LED framework or the PWM
> > > > > > framework.
> > > > > > 
> > > > > > A typical use case for the LED configuration is to drive RGB LEDs in
> > > > > > smartphones etc, for which the driver support multiple channels to be
> > > > > > ganged up to a MULTICOLOR LED. In this configuration the pattern
> > > > > > generators will be synchronized, to allow for multi-color patterns.
> > > > > > 
> > > > > > Signed-off-by: Bjorn Andersson <bjorn.andersson@...aro.org>
> > > > > > ---
> > > > > 
> > > > > Any feedback on this?
> > > > 
> > > > I asked in #linux-msm whether anything is wrong with the patterns,
> > > > since my Sony Discovery (sdm630 with a pm660l) blinks way quicker on a
> > > > pattern that's supposed to stay on for 1s and off for 1s:
> > > > 
> > > >     echo "0 1000 255 1000" > /sys/class/leds/rgb\:status/hw_pattern
> > > > 
> > > > It however seems to be broken in the same way on an older version now
> > > > (this might be v9 or v8) which I don't remember to be the case.  Can you
> > > > double-check if this is all working fine on your side?  If so, I'll have
> > > > to find some time to debug it on my end.
> > > > 
> > > 
> > > I had missed the fact that LPG_RAMP_DURATION_REG is two registers for
> > > msg and lsb, for a total of 9 bits of duration. So what you saw was
> > > probably ticking at 232ms.
> > > 
> > > Note though that the pattern uses the last time as "high pause", so I
> > > expect that you should have seen 232 ms of off, followed by 464ms of
> > > light.
> > 
> > Visual inspection seems to confirm those numbers indeed!
> > 
> > > I've fixed this for v11, both rejecting invalid input and writing out
> > > all 9 bits.
> > 
> > Doesn't that 512ms limit, together with using only the last value for
> > hi_pause (and not the first value for lo_pause) force users to write
> > patterns in a certain way which is not easily conveyed to the caller
> > except by reading the comment in the driver?  I'd guess lo_pause can be
> > used even if not in ping-pong mode, it should just hold at the first
> > value for the given duration?
> > 
> > (That said hw_pattern is anyway already riddled with device-specific
> > information, such as only having one `delta_t` which functions as the
> > step size for every entry, and with the change above would need to be
> > sourced from another step that's not the first.)
> > 
> 
> Perhaps we should clarify the single delta_t by requiring all those
> delta_t to be the same, rather than ignoring their value.
> 
> I.e. we make the ping-pong pattern:
> 
> <value> <lopause+t> ... <value[N/2-1]> <t> <value[N/2]> <hipause+t> <value[N/2-1]> <t> ... <value> <t>
> 
> And for non-ping-pong:
> 
> <value> <lopause+t> <value> <t> ... <value> <t> <value> <hipause + t>
> 
> 
> What do you think?

Seems like a good idea, though we'll have to be careful to communicate
this lopause+t value for the first entry and hipause+t for the
middle/last (through a dev_err I suppose) in case we reject values that
don't strictly adhere to this math.

> > Bit of a stretch, but perhaps worth noting anyway: should this be
> > written in documentation somewhere, together with pattern examples and
> > their desired outcome to function as testcases too?
> > 
> 
> There's a comment in lpg_pattern_set() where I tried to capture this.
> 
> I don't think it's worth documenting the behavior/structure away from
> the driver. But let's make sure it's captured properly there.

I've seen two other dirvers document the hw_pattern sysfs property under
Documentation/leds/.  Should be easier to find than a comment inside the
respective function deep in the kernel source tree I presume?

Quoting Documentation/ABI/testing/sysfs-class-led-trigger-pattern for
this sysfs property:

    Since different LED hardware can have different semantics of
    hardware patterns, each driver is expected to provide its own
    description for the hardware patterns in their documentation
    file at Documentation/leds/.

Doesn't need to be anything long, copying your inline comment would be a
great start.  Thanks!

- Marijn

Powered by blists - more mailing lists